Many prevention studies are now designed with complementary interventions in different settings.

Evaluations of these interventions require assessing the child's behavior in each of these settings.

Conducting these studies, therefore, may involve recruiting school districts, principals, classroom teachers, peers, parents, siblings, and in later years, employers and intimate partners.

These participants may be considered natural raters or satellite subjects, depending on their degree of involvement.

Issues of recruitment and retention thus are magnified in multimethod, multiagent studies.

To illustrate these issues, findings are presented for three studies conducted with risk populations in the past decade at the Oregon Social Learning Center : a passive longitudinal study, a selected prevention study, and an indicated prevention study.

Findings indicate that achieving high recruitment and retention rates for at-risk and high-risk subjects in multisetting studies is possible, and that a developmental approach should be taken to recruiting risk populations.